Emoji 18.0 is nineteen records and nine drawings nobody has drawn
Pick Emoji indexes 3,963 emoji records across eighteen releases. The newest of those releases, Emoji 18.0, accounts for nineteen of them. That is a small release by this dataset’s own standards — Emoji 4.0 contributed 598 records, Emoji 0.6 contributed 719, and Emoji 17.0 contributed 163 — but it is a useful one to look at closely, because it is the release where you can still watch every part of the pipeline arrive separately.
Nineteen records is not nineteen pictures
Nine of the nineteen are genuinely new characters:
| Character | Name | Codepoint |
|---|---|---|
| | Cracking face | U+1FAEB |
| | Leftwards thumb sign | U+1FAF9 |
| | Rightwards thumb sign | U+1FAFA |
| | Monarch butterfly | U+1FACC |
| | Pickle | U+1FADD |
| | Lighthouse | U+1F6D9 |
| | Meteor | U+1FA8B |
| | Eraser | U+1FA8C |
| | Net with handle | U+1FA8D |
If that first column is a column of empty rectangles on your screen, nothing is wrong with it. Your system font has no glyph for any of these yet, which is the same fact the rest of this post is about, arriving a few paragraphs early.
The remaining ten are skin-tone variants: the two thumb signs each take the five Fitzpatrick modifiers, which is five extra records apiece. That is why twelve of the nineteen sit in the people-and-body group, against two in travel-and-places, two in objects, and one each in smileys-and-emotion, animals-and-nature and food-and-drink. Two hand gestures dominate the release by record count without dominating it by drawing count.
The codepoints are not new territory
It is tempting to read a new emoji release as new ground being broken at the end of the encoding space. The codepoints say otherwise. Line the nine up against their neighbours in this dataset and they sit in holes inside blocks that were allocated years earlier:
- U+1FAEB Cracking face follows U+1FAEA, which this dataset records as Emoji 17.0’s distorted face. The rest of that run — U+1FAE0 through U+1FAE7 — is Emoji 14.0.
- U+1FA8B, U+1FA8C and U+1FA8D (meteor, eraser, net with handle) run straight on from U+1FA8A, Emoji 17.0’s trombone, in a block that starts with Emoji 12.0’s yo-yo at U+1FA80.
- U+1F6D9 Lighthouse follows U+1F6D8, Emoji 17.0’s landslide, in a block whose first entry here is Emoji 1.0’s place of worship at U+1F6D0.
- U+1FADD Pickle is the odd one: it fills a gap between two Emoji 16.0 characters, root vegetable at U+1FADC and splatter at U+1FADF.
Unicode does not renumber and does not compact. A codepoint’s numeric neighbours tell you about the block it was carved out of, never about when it arrived. If you want the release, read the release; this site puts it on every emoji page and gives each one its own version index.
Nobody has drawn them
This is the part that makes Emoji 18.0 worth a post. Pick Emoji shows every emoji across five openly licensed artwork sets, and measures coverage by counting the artwork files actually present for each one. The measurement:
- Twemoji — 3,944 of 3,963
- Noto Emoji — 3,944 of 3,963
- OpenMoji — 3,944 of 3,963
- Fluent Emoji — 3,145 of 3,963
- Blobmoji — 2,616 of 3,963
Three of the five are missing exactly nineteen records, and those nineteen are exactly Emoji 18.0. Not nineteen scattered gaps that happen to number nineteen — the same nineteen, in all three sets. Twemoji, Noto and OpenMoji have drawn every other emoji in the dataset and none of these. Fluent and Blobmoji have much larger gaps for unrelated reasons, and these nineteen fall inside them too.
So a page like pickle has nothing to show you but the character itself. It renders the raw Unicode glyph at a large size and says so plainly, rather than presenting five empty boxes.
The text data has not arrived either
The gap is not only visual, and that is easy to miss. Emoji carry a second layer of data that has nothing to do with artwork: the shortcodes platforms type them with, and the keywords search engines match them on. Across the dataset:
- 3,737 of 3,963 records carry a CLDR shortcode
- 1,699 carry a Slack shortcode
- 1,663 carry a GitHub shortcode
All nine new Emoji 18.0 characters carry none of the three. And of all 3,963
records in the dataset, exactly nineteen have an empty keyword list — the same
nineteen. There is no :pickle: yet, and searching this site for “gherkin”
will not find one, because no keyword source has published a keyword for it.
Why it works this way
A Unicode emoji release is a specification. It assigns a codepoint, fixes a name, and settles which group the character belongs to. It does not ship an image, a font, a shortcode or a keyword list, and it does not oblige anyone to produce them on any particular schedule. Each of those is separate work by a separate group: type designers at Google, the volunteers behind OpenMoji, the maintainer of Blobmoji, CLDR’s translators, whoever maintains a chat platform’s shortcode table.
A brand-new emoji is therefore a real character with a real codepoint that almost nothing can draw and almost nothing can search for. It is not broken. It is early. What changes first, usually, is artwork: the coverage numbers on the vendor pages move, and 3,944 becomes 3,963. The keyword and shortcode layer tends to lag further behind, because it has to be translated rather than merely drawn.
Those coverage counts are a snapshot of the artwork vendored into this site’s repository, not a live query. If the numbers above still read 3,944 some time from now, that means this site has not re-synced, not necessarily that the artwork does not exist yet.